Adaptive Vehicle Navigation Using Sparse Maps and User Overrides
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in navigating due to the sheer volume of data required for traditional mapping technologies, which can limit or adversely affect navigation, and the need for efficient data storage and processing of vast amounts of image, map, and sensor data.
Innovation Solution
The use of sparse maps and cameras for autonomous vehicle navigation, incorporating polynomial representations of target trajectories and landmarks, along with adaptive navigation systems that integrate user intervention and selective feedback, to reduce data storage and enhance navigation efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional mapping technology is used for autonomous vehicle navigation, then comprehensive navigation information is available, but data storage and processing requirements become excessively high
Solution Approach 1:
The patent extracts only the essential navigation elements (landmarks, road geometry, traffic signs) from the complete environmental data, creating a sparse map that contains only the information necessary for navigation. This extraction principle reduces data volume while maintaining navigation reliability by focusing on key features rather than storing all environmental details.
Solution Approach 2:
The navigation map is segmented into discrete landmarks and road segments rather than storing continuous complete environmental data. Each landmark is represented by key parameters (position, type, visual features) rather than full imagery, enabling efficient storage and processing while maintaining navigation accuracy through strategic selection of navigation-critical elements.
2Reliability
If complete environmental data is stored for navigation, then comprehensive situational awareness is achieved, but data processing time and computational load increase
Solution Approach 1:
The system extracts only navigation-relevant features from environmental data, such as landmark positions, road boundaries, and traffic sign locations, rather than processing complete image streams. This selective extraction maintains situational awareness for navigation purposes while dramatically reducing computational load and processing time.
Solution Approach 2:
The sparse map applies different levels of detail to different spatial locations based on navigation importance. High-detail representation is allocated to critical navigation elements (intersections, landmarks, traffic signals) while less critical areas use simplified representations, optimizing processing efficiency while maintaining situational awareness where it matters most.
Data Source
AI summary
Systems and methods are provided for autonomous navigation based on user intervention. In one implementation, a navigation system for a vehicle may include least one processor. The at least one processor may be programmed to receive images acquired by a camera from an environment of a vehicle; determine a navigational maneuver for the vehicle based on analysis of one or more of the plurality of images; cause the vehicle to initiate the navigational maneuver; receive a user input causing an override to alter the initiated navigational maneuver; determine navigational situation information relating to the vehicle via analysis of the images; determine, based on the navigational situation information, whether the user input is associated with a transient condition; and when the user input is not associated with a transient condition, store the navigational situation information in association with information relating to the user input.